I'm passionate about creating exciting and effective visualizations to disseminate my work. Here are two examples of such efforts:
I created this animation for a visualization competition at a Northwestern Computational Science Symposium where it was awarded first prize. It captures for a general audience how supernova discovery and classification workflows function and highlights how my BTSbot model enabled the world's first fully automated discovery and classification of a supernova.
The interactive panel below shows a 2-D view of BTSbot's learned representation of input ZTF data. You can pan, zoom, filter, and inspect individual sources to help understand what BTSbot is doing.
Tips: Use the top left drop down menu to color points by certain parameters like days_to_peak (an estimate of the source's rise time) or magpsf (the source's brightness in the given alert) to see correlations in the latent space. You can also select regions of the latent space with the rectangle and lasso tools in the bottom bright. For convenience, hide the side and bottom panels with the toggles in the top right corner.
Note: the interactive panel does not function on some browsers so try another browser if it doesn't load.